Abstract
In this paper we report our experience in solving min cost o w problems approximately by transforming them to network analysis problems. In the process we solve large (of the order of a million nodes) resistive networks. The preconditioned conjugate gradient (PCG) method appears the most suitable for this problem but runs into convergence diculties if the conductance values have the high range of 1 10 8 . We solve this problem by developing a variation of the PCG (which is described in the paper) and using it to solve hybrid analysis equations of the network. This suggests a relook at commonly used algorithms in computational linear algebra by associating an electrical network with the linear equations in question. In order to make the paper self contained we give a formal description of commonly used network analysis procedures such as nodal, loop and hybrid analysis.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call